The Benefits of Linking CGE and Microsimulation Models: Evidence

advertisement
DISCUSSION PAPER SERIES
IZA DP No. 3715
The Benefits of Linking CGE and Microsimulation
Models: Evidence from a Flat Tax Analysis
Andreas Peichl
September 2008
Forschungsinstitut
zur Zukunft der Arbeit
Institute for the Study
of Labor
The Benefits of Linking CGE and
Microsimulation Models:
Evidence from a Flat Tax Analysis
Andreas Peichl
University of Cologne
and IZA
Discussion Paper No. 3715
September 2008
IZA
P.O. Box 7240
53072 Bonn
Germany
Phone: +49-228-3894-0
Fax: +49-228-3894-180
E-mail: iza@iza.org
Any opinions expressed here are those of the author(s) and not those of IZA. Research published in
this series may include views on policy, but the institute itself takes no institutional policy positions.
The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center
and a place of communication between science, politics and business. IZA is an independent nonprofit
organization supported by Deutsche Post World Net. The center is associated with the University of
Bonn and offers a stimulating research environment through its international network, workshops and
conferences, data service, project support, research visits and doctoral program. IZA engages in (i)
original and internationally competitive research in all fields of labor economics, (ii) development of
policy concepts, and (iii) dissemination of research results and concepts to the interested public.
IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion.
Citation of such a paper should account for its provisional character. A revised version may be
available directly from the author.
IZA Discussion Paper No. 3715
September 2008
ABSTRACT
The Benefits of Linking CGE and Microsimulation Models:
Evidence from a Flat Tax Analysis*
Mircrosimulation models (MSM) and Computable General Equilibrium models (CGE) have
both been widely used in policy analysis. The combination of these two model types allows
the utilisation of the advantages of both types. The aim of this paper is to describe the stateof-the-art in simulation and to demonstrate the benefits of linking both model types modelling
flat tax reform proposals for Germany. Taking the general equilibrium effects into account has
important implications for the evaluation of a tax reform. The analysis shows that a personal
income flat tax can indeed overcome the fundamental equity efficiency trade-off in the longrun while simultaneously increasing the tax revenue. However, this result does not hold for a
flat tax combining a personal income flat tax with a corporate cash flow flat tax, even when
allowing for an ex-post loss in revenue.
JEL Classification:
Keywords:
D58, H2, J22
microsimulation, CGE, linked micro macro models, flat tax
Corresponding author:
Andreas Peichl
IZA
P.O. Box 7240
53072 Bonn
Germany
E-mail: peichl@iza.org
*
I would like to thank Christian Bergs, Markus Clauss, Clemens Fuest, Nicolas Hérault, Thilo
Schaefer, seminar participants in Cologne and Leuven and participants of the EcoMod conference in
Berlin for their helpful contributions. The usual disclaimer applies.
1
Introduction
The introduction of a ‡at tax system is widely seen as a reform which may boost e¢ ciency,
employment and growth through simpli…cation and higher incentives. However, inequality is
expected to increase as a consequence of a ‡at tax reform. In the discussion of the ‡at tax “a
notable and troubling feature [...] is that it has been marked more by rhetoric and assertion
than by analysis and evidence“.1 Given that ‡at taxes have not yet been implemented in
Western European countries, the e¤ects of ‡at tax reforms in these countries can only be studied
on the basis of simulation models. The method of simulation analysis aims at analysing and
quantifying the economic e¤ects of di¤erent policies based on the given institutional background
to compare and evaluate di¤erent reform proposals with respect to equity and e¢ ciency e¤ects.
For the analysis of …scal reforms, microsimulation (MSM) and computable general equilibrium (CGE) models have been widely used in the literature. CGE models consider various
interdependencies and facilitate simulating behavioural responses and adjustments on several
markets. In contrast, microsimulation models consider only the household side of the economy,
which allows for more heterogeneity and a much more detailed mapping of the complex tax
bene…t system. Combining these two model types enables the utilisation of the complementary
advantages. The aim of this paper is twofold. First, possibilities of linking microsimulation and
CGE models are shown. Then the bene…ts of linking both model types are illustrated using a
‡at tax example for Germany.
The most popular ‡at rate tax proposal is the ”Flat Tax“ of Hall and Rabushka (1985)
(HR), which combines a cash ‡ow taxation on corporate incomes with the same single marginal
tax rate on labour income. This proposal, however, has not been implemented in its pure form
in any country yet. Previous reforms considered a ‡at rate personal income tax as an indirect
progressive tax schedule with a basic tax allowance and a uniform marginal tax rate. In the
literature, there are several simulation studies on e¢ ciency and equity aspects of such (revenue
neutral) ‡at income tax reforms. One focus of these studies is the impact on employment and
growth.2 A second group of studies focuses on the distributional e¤ects of ‡at tax reforms.3
In summary, the main result of this literature is that a ‡at tax can increase the e¢ ciency in
terms of employment and growth but will most likely lead to redistribution in favour of high
income households. All previous studies support the existence of a trade-o¤ between equity
and e¢ ciency, i.e. either inequality and employment both increase or decrease but it is not
1
Keen et al. (2007), p. 3.
See e.g. Browning and Browning (1985) for the U.S., Heer and Trede (2003) for Germany, Cajner et al.
(2006) for Slovenia or Jacobs et al. (2007) for the Netherlands.
3
See e.g. Caminada and Goudswaard (2001) for the Netherlands, Aaberge et al. (2000) for Italy, Norway
and Sweden, Decoster and Orsini (2007) for Belgium, González-Torrabadella and Pijoan-Mas (2006) for Spain,
Fuest et al. (2007b) for Germany and Paulus and Peichl (2008) for a cross country study of 10 EU countries.
2
1
possible to decrease inequality while increasing employment. However, a HR-type ‡at tax may
lead to further e¢ ciency gains due to more investment and labour demand as it is also likely to
reduce tax distortions in the corporate sector. These outcomes, however, can only be analysed
taking the general equilibrium e¤ects into account.
The introduction of a HR-type ‡at tax reform has been analysed before, especially for the
U.S. using CGE models. Browning and Browning (1985) estimate an increase in labour supply
by 5%. Stokey and Rebelo (1995) compare and summarise di¤erent studies for the U.S. and
conclude that a ‡at tax reform would have little e¤ect on the growth rate. Gale et al. (1996)
analyse the e¤ects of introducing a HR ‡at tax in the U.S.. They conclude that high income
households pro…t most while households with low incomes su¤er from a ‡at tax reform. Ho and
Stiroh (1998) and Dunbar and Pogue (1998) show that high income households gain whereas
especially middle income households are burdened by a ‡at tax reform. Ventura (1999) …nds
an increase in capital accumulation and a redistribution in working hours and income in favour
of the top of the distribution. Altig et al. (2001) …nd that output, labour supply and wages
increase and that the lowest income households lose through a ‡at tax. Cassou and Lansing
(2004) …nd that a ‡at tax reduces growth in the short run if revenue-neutrality is maintained,
but increases capital accumulation and therefore growth in the long run. Díaz Giménez and
Pijoan-Mas (2006) analyse two di¤erent ‡at tax proposals for the U.S. and …nd that the reform
with the lower (higher) marginal rate increases (decreases) output and inequality, but decreases
(increases) aggregate welfare. However, in both scenarios the poor obtain signi…cant welfare
gains. Nielsen et al. (1999) …nd signi…cant e¢ ciency gains but negative distributional e¤ects
for a ‡at tax in Denmark. However, none of these studies uses a linked MSM-CGE model.
In this paper, we analyse a cash ‡ow ‡at tax reform of the German corporate and personal
income tax system according to the proposal by Mitschke (2004), which is closely related
to the Hall-Rabushka idea. Our focus lies on the e¤ects on tax revenue, income distribution,
employment and economic growth. Our analysis is based on a simulation model for the German
tax and transfer system (FiFoSiM) using income tax micro data and household survey data. To
be able to simulate equity and e¢ ciency e¤ects within the same microeconometric framework,
we use the linked MSM-CGE module of FiFoSiM. With its socio-economic and demographic
structure, Germany can be seen as a typical Western European democracy. Therefore, the
qualitative results of our analysis should be of interest to a wider range of countries.4
Our analysis yields the following results: Applying the linked model to a not revenue neutral
‡at tax proposal shows that taking the general equilibrium e¤ects into account indeed increases
the expected e¢ ciency gains in the long-run. The overall employment e¤ects are larger than the
4
It has to be taken into account, though, that the structures of the tax bene…t systems do vary considerably
among the countries of Western Europe.
2
labour supply reactions because of reduced costs of labour and capital resulting in increasing
labour and investment demand. Therefore, a personal ‡at income tax can indeed overcome
the fundamental equity-e¢ ciency trade-o¤. However, combining this ‡at tax with a cash ‡ow
‡at tax on business income still increases inequality due to the large gains at the top of the
distribution. Therefore, due to their limited e¢ ciency e¤ects and their problematic short-term
distributional impact, ‡at tax reforms are unlikely to spill over to the grown-up democracies of
Western Europe.
The setup of this paper is as follows. Section 2 describes microsimulation models (MSM),
computable general equilibrium models (CGE) and methods to link them. Section 3 brie‡y
discusses potential problems of simulation studies. Section 4 describes the microsimulation and
CGE model used for the application. In section 5, the ‡at tax reform proposal for Germany is
decribed, which is analysed in section 6. Section 7 concludes.
2
Simulation models
In this section, simulation models are introduced and described as a method for the ex-ante
evaluation of the various consequences of …scal reforms. Various proposals to reform the complex
tax and bene…t systems exist in every welfare state, and new proposals are being presented in
policy debates each year. In the run-up of the implementation of a certain reform, in many cases
the expected consequences remain an unsolved puzzle. Especially the behavioural responses
of the (bounded) rational individuals are extremely di¢ cult to estimate ex-ante. Knowing
these responses, though, helps to evaluate and judge di¤erent reform proposals regarding their
target achievement and cost e¢ ciency. Estimating them, however, is not a trivial task. The
complexity of existing welfare states requires the usage of simpli…ed models for the evaluation of
reform proposals. Theoretical models should be kept small and simple to be able to understand
the general principles. They allow to point out a single argument in a simpli…ed framework
and to construct hypotheses which can be tested empirically. Empirical models allow for an
econometric evaluation of a given reform and are especially useful whenever the magnitude, and
thus not only the sign, of the e¤ects are to be estimated. The quality of the empirical analysis
crucially depends on the availability of high quality micro data. If the reform already has been
implemented and data is available for the pre- and post-reform period, an ex-post analysis
is possible using standard econometric procedures. On the other hand, if the reform has not
been implemented, only simulation models can provide information for an ex-ante analysis of
di¤erent reform proposals.
In general, simulation models are tools which are designed to answer “what if” questions
about di¤erent policy reform options. The method of simulation analysis uses actual economic
3
data to estimate how an economy might react to changes in external factors (e.g. policy
parameters). Simulation models map the complex tax bene…t system to analyse and quantify
the e¤ects of di¤erent policies based on the given institutional background. Other than in
the natural sciences, it is seldom possible in economics to construct natural experiments for
the analysis of a given treatment (policy). Policy simulations can be interpreted as quasiexperiments which allow the economist to ex-ante analyse a reform proposal controlling for
behavioural responses of the micro units in the economy. Simulation models are frequently
used by economists, policy-consultants and policy-makers to predict the impacts of changes in
…scal policies on individuals (gains and losses, income distribution), the government budget and
key economic indicators (e.g. growth, employment, prices, consumption) to provide the political
decision makers with well-founded decision guidance. This is done by setting up alternative
scenarios by varying the rules of the tax bene…t system and then simulating the impacts of
these changes on individual and aggregated variables.
Several di¤erent types of simulation models can be used depending on the research question
in mind. In the following, we focus on public economics models which can be used for the
analysis of tax bene…t reforms. The aim of this section is to describe the status-quo of the
research in the …eld of simulation models by introducing and comparing the di¤erent types of
models. The following subsections describe in more detail the standard procedures of computable general equilibrium models (CGE), microsimulation models (MSM) and linked micro
macro models.
2.1
Computable General Equilibrium Models (CGE)
General equilibrium theory has provided important insights about mechanisms that determine
the allocation of resources on mutually interdependent markets. Computable General Equilibrium (CGE)5 models use this general equilibrium theory to empirically analyse and quantify
this allocation of resources. CGE models are used instead of analytical general equilibrium
models whenever the size and complexity make such analytical models mathematically intractable. Although CGE models are based on the microeconomic general equilibrium theory they
usually use aggregated macro data for the analysis. The …rst CGE model was presented by
Johansen (1960). With the development of fast computers and suitable software a large number
of CGE models has been developed and applied to policy analysis since then.
CGE models use as realistic values as possible of exogenous variables (e.g. elasticities, tax
rates) to numerically compute the values of the endogenous variables (e.g. prices, quantities)
5
Sometimes this class of numerical economic models is also called Applied General Equilibrium (AGE)
models. This subsection is based on Bergs and Peichl (2008). Further introductions to CGE models can be
found in Kehoe and Prescott (1995), Fehr and Wiegard (1996) or Bergman (2005).
4
with the aim of quantifying economic equilibria to compare the impact of policy measures
on these equilibria. Applications of CGE models include analyses of tax reforms, changes
in trade policy regimes, economic integration, agricultural policies and energy policies. The
analysis focuses particularly on the long-run allocation of factors and goods, whereas shortterm distributional e¤ects cannot be analysed in a sophisticated way using this type of models.
2.1.1
Standard procedure
A CGE model is usually a multi-sector model based on real world data of one or several national economies to model the interactions of individual households and …rms on interdependent
markets. However, in a typical CGE model there is only one or possibly a few representative
agents, while the number of …rms (production sectors) is generally larger. A CGE model consists of equations describing the variables and a database consistent with these equations. For
all agents (households, …rms, government) an optimising behaviour, i.e. utility and pro…t maximization on the part of households and …rms, is assumed to model their behaviour on di¤erent
markets. In general, standard models assume product and factor markets to be competitive
and relative prices ‡exible enough to simultaneously clear all markets. However, it is possible to
allow for non-market clearing (e.g. unemployment or inventories), imperfect competition (e.g.
monopolistic competition)6 , heterogeneous agents, and taxes or externalities (e.g. pollution).
CGE models are almost always focused on the real side of the economy and thus do not include
…nancial assets. Consequently a typical CGE model endogenously determines relative product
and factor prices, but cannot determine nominal prices. In particular CGE models are aimed
at quantifying the impact of speci…c policies on the equilibrium allocation of resources and
relative prices of goods and factors.
For the numerical computation of equilibria, it is essential to specify functional forms of
production and utility functions as well as the values of the exogenous parameters of the model.
The speci…cation of these functions and parameters is of key importance for the model results.
There are two general approaches. On the one hand, these parameters can be estimated using
econometric methods based on time series or panel data, or, on the other hand, these values
can be calibrated7 using a micro consistent dataset re‡ecting an economic equilibrium at a
given point in time. Often both methods are combined and some parameters are estimated (or
estimates are taken from the literature) and other parameters are calibrated to replicate the
benchmark equilibrium given in the data. The construction of a micro consistent database can
be rather time consuming if information from several sources has to be combined. In general,
the skeletal structure is based on an input-output-table which is enhanced to a so called “Social
6
7
See e.g. Harris (1984).
See Mansur and Whalley (1984) and Lau (1984) for an extensive discussion of the calibration method.
5
Accounting Matrix“ (SAM).8
CGE models allow quantifying the impacts of policy reforms. However, they - as all models
do - obviously rest upon strong simplifying assumptions about optimising behaviour, competitive markets and ‡exible prices. Furthermore, the calibration method which is often used to
de…ne key model parameters is often seen to be rather arbitrary. In view of this, the validity and
therefore also the usefulness for policy evaluation of the results is often seriously questioned.
However, the usefulness of a CGE model depends on the aims and purposes it was designed for
and what the alternatives are. If a general equilibrium model cannot be solved analytically, a
numerical solution can help to identify general equilibrium e¤ects of policy changes even if key
parameters of the model are quite uncertain. The role of these parameters for the results can
be explained using extensive sensitivity analyses. Furthermore, even if the precise magnitude
of the e¤ects remains uncertain, it still might be possible to identify if the e¤ects are small or
large or at least to compare and rank di¤erent scenarios based on these results.
2.1.2
CGE categories
Di¤erent categories of CGE models can be distinguished. First of all, it should be emphasised
that the models within each category can di¤er in many ways (e.g. heterogeneity in terms of
number of agents, sectors, factors or commodities, as well as the representation of the government or foreign trade with the rest of the world). Nevertheless, some common elements can
be attributed to di¤erent categories. The most appropriate classi…cation for most modelling
approaches is to distinguish between static and dynamic models. In addition it is useful to
distinguish between single-country (national) and multi-country (global) models. Generally,
single-country models are much more detailed in terms of sectors and household types and are
especially designed to analyse country-speci…c policy issues. Multi-country models are used for
multi-lateral policies (e.g. trade agreements, emission trading schemes) and are usually less
detailed in terms of intra country heterogeneity.
Comparative-static models are by far the most common class of CGE models.9 The economy is modelled at two given points in time only: the status quo benchmark and the future
counterfactual equilibrium. The results of these models are the long-run di¤erences (usually
reported in percent changes) between the benchmark equilibrium and the future equilibrium to
which the economy converges after a given exogenous shock. The transition path towards this
new equilibrium is not explicitly modelled. This, however, allows for a more detailed speci…cation of the single-period economy in terms of numbers of agents, sectors or commodities to be
8
See Pyatt and Round (1985) and Kehoe (1998) on how to construct a SAM.
See e.g. Shoven and Whalley (1984), Shoven and Whalley (1992) or Kehoe and Kehoe (1994) for an
introduction.
9
6
represented in the model.
Dynamic CGE models, by contrast, explicitly model this transition path. The equilibrium
is computed for all future points in time (at discrete intervals) until the model horizon ends.
These models are far more challenging to design, maintain and solve but allow a more realistic
representation of the adjustment process of a policy change. However, the increasing complexity
of dynamic models often reduces the heterogeneity of the agents and therefore, in general,
smaller number of agents and sectors are represented in dynamic models. Dynamic models
assume rational expectations of agents, i.e. they use all available information for the best
guess of the future. This makes it necessary to simultaneously solve for all periods. Stochastic
CGE models explicitly incorporate uncertainty about the future into the analysis. In contrast,
recursive-dynamic CGE models assume that behaviour depends only on current and past states
of the economy (assuming myopic expectations).10 These models can be solved sequentially (one
period at a time). Dynamic models can be further distinguished according to the representation
of the household side of the economy. CGE models based on the Ramsey (1928) growth model
assume an in…nitely living representative agent. In contrast, overlapping generations (OLG)
models assume periodically overlapping generations. These models can be further distinguished
into models with identical lifetime for all generations (see Auerbach et al. (1983) and Auerbach
and Kotliko¤ (1987)) or models with a stochastic time of death (see Blanchard (1985)).
2.1.3
Applications
There are many …elds for applications of CGE models in economics. The short review in this
section focuses on recent models for the analysis of tax reforms especially in Germany.11
Hutton and Ruocco (1999) use a static model to analyse di¤erent tax reforms in Germany,
France, Italy and the UK especially with respect to their labour market performance. Sørensen
(2002) also uses a static model for the analysis of the German tax reform of 2000.
Keuschnigg and Dietz (2003) use an OLG model to analyse the corporate tax reform in
Switzerland. This model is adapted to Germany by Radulescu and Stimmelmayr (2004) who
analyse the introduction of a dual income tax. Fehr and Wiegard (1998) use an AuerbachKotliko¤ OLG model to analyse di¤erent income tax reform proposals.
Graa‡and et al. (2001) and Bovenberg (2003) use the recursive-dynamic model MIMIC to
analyse the impact of tax reforms on the Dutch labour market. The German PACE-L model was
constructed following the MIMIC approach. This model is used to analyse the labour market
e¤ects of hypothetical income tax (Boeters, Gürtzgen and Schnabel (2006)), VAT (Boeters,
10
See Ballard et al. (1985).
See e.g. Fehr and Wiegard (1996) for a review of additional models. An application for the US can for
example be found in Altig et al. (2001).
11
7
Böhringer, Büttner and Kraus (2006)) or welfare (Böhringer et al. (2005)) reforms.
2.2
Microsimulation Models
Microsimulation models (MSM) are quantitative models of the tax bene…t system and have
been introduced into social sciences and economics by Orcutt (1957) and Orcutt et al. (1961).12
2.2.1
Standard procedure
Microsimulation models are partial equilibrium models focusing on one side (e.g. the household
side) of markets and do not consider the broader economic environment in which the micro units
are acting. Tax bene…t models usually only simulate …rst round e¤ects where the behavioural
response of the agents (to the change imposed on the system) is not captured. These tax bene…t
models can be combined with a labour supply model allowing the simulation of (partial) second
round e¤ects in terms of behavioural responses on the labour market.
In the centre of this microanalytic approach is the behaviour of individual agents to which
the observed social and economic processes can be attributed. MSM are based on micro data
which o¤er a great ‡exibility especially regarding a detailed mapping of the complex tax bene…t
system. These data samples allow the modelling of structural characteristics of micro units
(persons, households, …rms) within a particular tax bene…t system. This allows MSM to re‡ect
the considerable heterogeneity within the population by taking into account the characteristics
and circumstances of each individual. MSM can be used to analyse the status quo of tax
bene…t rules by providing detailed data on single policy variables. Furthermore, these factuals
are then used as a benchmark (baseline) to evaluate the impact of a reform scenario. Such a
counterfactual can be a single policy or various interacting and interdependent policies within
the complex tax bene…t system as well as fundamental reform proposals of the whole system.
To analyse a reform scenario, the socioeconomic system is modelled and applied to the sample
micro units, which in turn are weighted with population weights to extrapolate results for the
whole population. Therefore, the results of MSM can be analysed either at the individual
level or at di¤erent levels of aggregation. Despite these advantages there are some limitations.
Every empirical analysis relies on high quality data of all key variables. If such data is not
available, simplifying assumptions are necessary which lead to biased estimations. If, e.g., key
information necessary for some (minor) tax rules is not available, it is not possible to include
these rules into the mapping of the tax bene…t system.
12
This subsection is based on Peichl (2005). See e.g. Gupta and Kapur (2000), Harding (1996) or Bourguignon
and Spadaro (2006) for further MSM surveys.
8
2.2.2
MSM categories
Microsimulation models can be di¤erentiated according to the time dimension into static and
dynamic models and according to the modelling of behavioural responses into behavioural and
non-behavioural models.
2.2.3
Static vs. dynamic MSM
Static MSM use cross-sectional data at a given point in time. Often this data has to be aged to
the year of analysis which might be further in the future using the static ageing technique, i.e.
reweighting of individual records based on macroeconomic indicators.13 Static models mimic
the tax law by applying the (current or an alternative set of) tax bene…t rules to individual
units. These models are essential tax bene…t calculators for all individuals and therefore allow
to model the tax bene…t rules for every point of the income distribution. This allows the user to
simulate the instantaneous …rst-round e¤ects (in terms of the …scal and distributional e¤ects,
i.e. the gains and losses in di¤erent variables at the individual or aggregated level) of policy
changes. They allow for a comparative-static analysis of the pre- and post-reform state of the
economy without looking at the adjustment process.
Dynamic MSM try to endogenously explain this process of adaptation through the incorporation of dynamic ageing of individual records over time based on the probabilities of the
happening of di¤erent real life events (e.g. marriage, divorce, birth of a child). The relevant
life processes are simulated and the individual characteristics are recalculated at each period
in time which allows moving the micro units forward through time. On the one hand, dynamic MSM allows the modelling of demographic changes over time, but on the other hand,
dynamic models have a higher demand regarding the modelling, the data requirements and the
computational resources than static models. Therefore, often static models which are easier to
build and maintain are used in combination with a behavioural model for the analysis of the
short-term e¤ects of policy reforms.
2.2.4
Behavioural responses
Behavioural responses can be simulated with both static and dynamic models. Sometimes
behavioural models are labelled dynamic. Although dynamic models often include behavioural
responses, they do not necessarily have to include them.
Non-behavioural models do not allow the individuals to change their behaviour as a consequence of a given policy reform. These models are often used to estimate the immediate
…scal and distributional e¤ects. This is done by generating income pro…les for various groups
13
See Gupta and Kapur (2000) for a review of the ageing techniques.
9
of individuals to highlight discontinuities in the tax bene…t rules which in turn can be modi…ed
by policy-makers.
Behavioural models simulate some kind of behavioural response to a policy change. These
responses can include the supply and demand of factors and goods. The most common applications are models of labour supply. Labour supply models allow the modelling of both the
extensive (participation) and the intensive (hours worked) labour supply decision. The labour
supply model can be either integrated into the microsimulation model or it can be linked to a
MSM as an external module. There are several possibilities of how to model the labour supply
of a tax unit (e.g. individual vs. household labour supply, discrete vs. continuous working
hours, selection of utility functions). Recent surveys of the empirical labour market literature
and di¤erent kinds of labour supply models are for example provided by Heckman (1993),
Blundell and MaCurdy (1999) or Creedy et al. (2002).
2.2.5
Applications
Various microsimulation models exist worldwide. The review in this subsection focuses on
German models which are used for the analysis of reforms of the tax bene…t system.14
Wagenhals (2001) uses GMOD to simulate the …scal and distributional e¤ects of the income
tax reform act of 2000. Haan and Steiner (2005) use STSM to estimate the …scal, distributional
and labour supply e¤ects of the income tax reform act of 2000. Steiner and Wrohlich (2004)
analyse the labour supply e¤ects of the German system of family taxation. Merz and Zwick
(2002) and Merz et al. (2002) analyse the income tax reform act of 2000 with MICSIM especially
with respect to the upper end of the income distribution.
Immervoll et al. (2007) use the European model EUROMOD to analyse di¤erent reform
scenarios in the EU-15 countries and Paulus and Peichl (2008) analyse and compare the distributional and incentive e¤ects of di¤erent ‡at tax reforms in these countries.
2.3
Linked Micro-Macro Models
2.3.1
Standard procedure
On the one hand, CGE models provide an economy wide perspective of a given shock after
the economy has fully equilibrated. The need to specify and calibrate functional forms and
parameters for all agents on all markets reduces the number of agents to be modelled dramatically. Therefore, in general, few representative agents are used which reduces ‡exibility
and the possibilities of a detailed modelling of the tax-bene…t system. In consequence, these
models give no insight into how aggregate changes in the economy and the new equilibrium
14
See e.g. Wagenhals (2004) or O’Hare and Gupta (2000) for further surveys.
10
solution a¤ect individuals. On the other hand, MSM are based on micro data which o¤ers great
‡exibility speci…cally regarding the mapping of the complex tax bene…t system. These models
are partial equilibrium models focusing on one side (e.g. the household side) of markets and
do not consider the broader economic environment in which the micro units are acting.
During the last years, a tendency of linking micro and macro models has emerged in computational economics to utilise the complementary advantages of MSM and CGE models.15 A
linked model can provide a more powerful tool for policy analysis than using results from two
stand-alone MSM and CGE models.16 Outputs from the macro model can be used to align the
predictions of the micro model and to enable general equilibrium feedbacks and interactions
among variables in the micro model. Outputs from the micro model can be used to calibrate
the macro model and provide a microeconomic basis for aggregate behaviour. Hence, the key
advantage of a linked micro macro model is the feedback which is used to resolve the model
corresponding to a revised set of parameters. However, achieving these feedback e¤ects through
linking MSM and CGE models is not a trivial task.
The idea of linking micro- and macroeconomic simulation models is almost as old as the
stand-alone models themselves.17 Orcutt (1967) suggests to link models operating at di¤erent
levels of aggregation through intermediate variables. Nevertheless, the number of researchers
developing linked micro macro models is still very small worldwide (see Davies (2004)). Nonetheless, recent advances in computational and econometric methods are leading to a growing
interest in combining these modelling techniques. The yet recent development in this area can
be clearly attributed to the progress in computer and information technology which makes these
large-scale models feasible to solve.
There are two general possibilities for linking the models. On the one hand, one can completely integrate both models into a joint model18 or on the other hand, one could combine two
separated models via interfaces (layered approach).19 The …rst approach requires the complete
micro model to be included in the CGE model which demands high standards for the database and the construction of the integrated model. This often results in various simplifying
assumptions.
The layered approach can be di¤erentiated into “top-down“ (see Figure 1, left-hand side),
“bottom-up“ (see Figure 1, right-hand side) or “top-down bottom-up“ (see Figure 2) ap15
Cf. Davies (2004) for an overview. Most of these models deal with trade liberalization in developing
countries.
16
Cf. Anderson (1990).
17
See Orcutt (1967) or Conrad (1991) for Germany.
18
Cf. Cogneau and Robilliard (2000) or Cororaton et al. (2005). See also Arntz, Boeters and Gürtzgen
(2006) for description of how to integrate a discrete choice labour supply into a CGE model.
19
Cf. Bourguignon et al. (2003).
11
Figure 1: Top-down and bottom-up
proaches.20 The top-down approach computes the macroeconomic variables (price level, growth
rates) in a CGE model as input for the micro model which is adjusted to match an exogenous
macro aggregate. The bottom-up approach works the other way around and information from
the micro model (elasticities, tax rates) is used in the macro model (e.g. for calibration of
the representative agents). Both approaches su¤er from the drawback that not all feedback
is used. The top-down bottom-up approach combines both methods to a recursive approach.
In an iterative process one model is solved, information is sent to the other model, which is
solved and gives feedback to the …rst model. This iterative process continues until the two
models converge. Böhringer and Rutherford (2006) describe an algorithm for the sequential
calibration of a CGE model to use the top-down bottom-up approach with micro models with
large numbers of households.
20
Cf. Savard (2003) or Böhringer and Rutherford (2006).
12
Figure 2: Top-down bottom-up
2.3.2
Applications
So far, the application of linked micro macro models to analyse tax reforms is rather limited.21
Boeters et al. (2005) use the bottom-up approach to calibrate the three representative households of a CGE model to analyse di¤erent hypothetical reform proposals of the social assistance
bene…t system in Germany.22 A similar approach is chosen by Fuest et al. (2005a) to evaluate a tax reform proposal for Germany with respect to …scal, employment and growth e¤ects.
Arntz, Boeters, Gürtzgen and Schubert (2006) use the recursive top-down bottom-up approach
to analyse reform proposals designed to encourage labour supply at the lower end of the wage
distribution in Germany. However, various simplifying assumptions regarding the aggregation
and disaggregation of information on labour supply responses that is passed between the models
have to be made.
Aaberge et al. (2007) use an integrated micro macro model of Norway to analyse the impact
of population ageing on …scal sustainability with endogenous labour supply. Although their
model is integrated, they use an iterative approach. Labour supply responses are computed
using the MSM model and are then used in the CGE model to estimate changes in wage rates.
Rutherford et al. (2005) link a CGE model to the Russian Household Budget Survey (representing 55,000 households) in order to analyse the distributional e¤ects of Russia’s WTO
accession. They use the information of the micro model to calibrate the representative agent
of the CGE model and iterate both models until they converge in terms of price changes and
21
There are, however, more applications to trade reforms (see e.g. Davies (2004) for a survey). More recent
applications include Hérault (2007), who uses the top-down approach to analyse the e¤ect of trade liberalisation
on poverty in South Africa.
22
The same model and approach is also used by Boeters, Gürtzgen and Schnabel (2006).
13
aggregate demand equalling aggregate supply.
2.4
Summarising comparison of model types
The method of simulation analysis can be seen as an economic quasi-experiment for the ex-ante
evaluation of policy reforms to analyse and compare the impacts of di¤erent reform scenarios
before they are implemented in real life. Di¤erent types of simulation models evolved over time.
The three most common and appropriate models for the analysis of …scal reforms, microsimulation models (MSM), computable general equilibrium models (CGE) and linked micro macro
models have been described in this section. The main research questions for these categories
are presented in Table 1.
Framework
Data
Research questions:
- Growth
- Allocation / E¢ ciency
- Labour supply
- Labour demand
- Revenue
- Distribution
CGE MSM Micro-Macro
total
partial total
macro micro both
X
X
(X)
X
(X)
X
X
X
X
X
X
X
X
X
Table 1: Comparison of model types
CGE models excel through their outstanding theoretical foundation and the consideration
of various interdependencies. They allow estimating various behavioural responses and adjustments on several markets, e.g. modelling labour supply and demand on the labour market. In
contrast, microsimulation models take only the labour supply side into account. Nevertheless,
these models allow for a much more detailed mapping of the complex rules of the tax bene…t
system and account for a much greater heterogeneity than CGE models could ever do because
of the aggregated data they are based on. Furthermore, the need to specify and calibrate functional forms and key parameters gives rise to various critiques against CGE models because
of limited econometric foundations of the calibration technique in general. During the last
years, a tendency of linking micro and macro models has emerged to utilise the complementary
advantages of MSM and CGE models.
So far, the linked micro macro models which have been used for the analysis of tax bene…t
reforms do not su¢ ciently use all the possibilities stand-alone MSM and CGE models o¤er. The
further development of computational power and more powerful algorithms (see e.g. Rausch
and Rutherford (2007)) should allow for a complete integration of both types of models. This in
14
turn will then enable to analyse the complex interdependence of various policy measures with
respect to …scal, distributional, employment and growth e¤ects within the same econometric
framework. Further development on the microsimulation modelling of the corporate sector is
also desirable. This, of course, crucially depends on the availability of corporate micro data.
Such a corporate MSM could then be linked with a household MSM and integrated into a CGE
model which would then be based on micro data for both sides of the economy.
3
Potential problems
When conducting a simulation analysis or interpreting its results one should be aware of potential errors or biases. According to Betson (1990) “sampling errors“, “imputation errors“,
“ageing errors“, “individual response errors“ and “environmental errors“ can be distinguished
in general. Sampling errors can always occur when subsamples from the whole population are
used in an empirical analysis which can increase or decrease the variation in the data. Thus,
estimates from the simulation model might di¤er from estimation based on the whole population. If the sample was not drawn randomly, the assumptions of statistical procedures might
be violated and special corrections have to be used in the analysis. Furthermore, there might
be also “non-sampling errors“ in the dataset, resulting e.g. from non-response and reporting or
data processing mistakes. The weighting of the individual records with population weights to
estimate aggregated values for the whole population can reduce these errors, as the weighting
factors are chosen to ensure that the sample estimates conform to macroeconomic indicators
of the whole population. However, when these population factors are modi…ed, e.g. using
static ageing techniques of reweighting, this can give rise to ageing errors if the modi…cation
itself is biased. Imputation errors arise when data from di¤erent sources are used for the imputation of missing values or variables. As a consequence, distributional assumptions might
be violated leading to biased estimations. But not only the data is error-prone but also the
modelling of the benchmark or the counterfactual scenario itself gives rise to potential mistakes.
Individual response errors can arise from simplifying and/or behavioural assumptions in the
model. Simplifying assumptions are always subject to errors, but have to be used to overcome
data limitations or to make the model operational. Behavioural assumptions are necessary for
the estimation of behavioural responses. To do so, functional forms and co-variables of the
econometric model have to be speci…ed based on beliefs of the underlying behaviour of the
individuals. Incomplete or imperfect beliefs can lead to misspeci…cations and biased results.
Environmental errors can e.g. arise from the negligence of the broader economic environment
or individual reactions to policy changes.
When conducting a micro-macro linkage, several speci…c potential problems arise. To be
15
able to successfully link MSM and CGE models there have to be some common variables
through which the two models can exchange information. Usually, it is necessary to aggregate
or disaggregate these variables to be comparable with the variables in the other model. Of
course, the less variables have to be (dis)aggregated the more of the underlying heterogeneity
in the data will be retained. Furthermore, it has to be checked if the same variable in both
models represents the same population (e.g. household consumption in the micro model vs.
aggregated total consumption including government in the macro model). Functional forms
(e.g. the preference functions in the labour supply model and the aggregated utility in the
CGE model) have to be speci…ed in a consistent way.23 In addition, it has to be checked if
one run of each model represents the same time horizon. Usually, a MSM computes short-term
e¤ects, whereas a CGE models aims at the long-run equilibrium.24 Therefore, the information
exchange between both models has to take into account these temporal di¤erences. If, for
example, a given labour supply shock from the MSM model has not fully equilibrated within
the CGE model and information is passed back and forth again, a second labour supply shock
might overlap the …rst one and either foil or fortify the …rst shock.
When building and using a simulation model, a researcher should be aware of these potential
errors and should try to avoid them if possible or at least to document the possible biases in
the analysis. Extensive sensitivity analyses should be conducted when building a model or
simulating a new scenario. When interpreting the results of a simulation study, one has to be
aware of these potential errors and has to take a closer look at the underlying data, methods
and assumptions. Furthermore, estimations from simulation models should not be used as an
exact forecast of a single number but to compare and rank di¤erent scenarios according to
various dimensions. Despite all these potential errors, simulation models nevertheless provide
a powerful tool for the ex-ante evaluation of …scal policy reform proposals.
4
Database and model
Our analysis is based on a behavioural simulation model for the German tax and transfer system
(FiFoSiM) using income tax and household survey microdata.25 The approach of FiFoSiM is
23
Verboven (1996) shows that a nested logit speci…cation of the individual direct utility function (which is
usually used in labour supply models) can be aggregated to a global CES utility function (which is usually used
in CGE models).
24
This is, of course, not always the case. The temporal horizon of CGE models depends on the closure
conditions. Particularly the closures for the capital market can be designed such that the CGE model can be
used for short or medium term analyses with a …xed stock of capital not mobile across sectors.
25
This section is based on the English documentation of FiFoSiM (see Peichl and Schaefer (2006)), which is
a short version of the detailed German description (see Fuest et al. (2005b)). Due to a lack of space, we are not
able to report the equations underlying the formal structure of the model in this paper. They can be, however,
found in the model documentation. See also www.cpe-cologne.de for further information.
16
innovative insofar as it creates a dual database using two micro data sets for Germany: FAST01
and GSOEP.26 FAST01 is a microdataset from the German federal income tax statistics 2001
containing the relevant income tax data of nearly 3 million households in Germany. Our
second data source, the German Socio-Economic Panel (GSOEP), is a representative panel
study of private households in Germany. The simultaneous use of both databases allows for
the imputation of missing values or variables in the other dataset using techniques of statistical
matching.
Figure 3: Basic setup FiFoSiM
Figure 3 shows the basic setup of FiFoSiM. The layout of the tax bene…t module follows
26
In the last years several tax bene…t microsimulation models for Germany have been developed (see for
example Peichl (2005) or Wagenhals (2004)). Most of these models use either GSOEP or FAST data. FiFoSiM
is so far the …rst model to combine these two databases.
17
several steps: First, the database is updated using the static ageing technique which allows
controlling for changes in global structural variables (through reweighting of the sample) and a
di¤erentiated adjustment for di¤erent income components of the households (through uprating
of various income components). Second, we simulate the tax and bene…t system in 2008 using
the uprated data. This allows us to compute the disposable incomes for each person and
household taking into account the detailed rules of the complex tax bene…t system. The basic
steps for the calculation of the personal income tax under German tax law are as follows.
The income of a taxpayer from di¤erent sources is allocated to the seven forms of income
de…ned in the German income tax law. For each type of income, the tax law allows for certain
speci…c income related expenses. Then, general deductions like contributions to pension plans
or charitable donations are taken into account and subtracted from the sum of incomes, which
gives taxable income as a result. Finally, the income tax is calculated by applying the tax rate
schedule to taxable income. To derive the disposable income Y from gross income G; received
bene…ts (like unemployment bene…t, social assistance, child bene…ts, etc.) are added and taxes
T and social insurance contributions S are subtracted:
Y =G+B
T
S
Third, the individual results are multiplied by individual sample weights to extrapolate the
…scal e¤ects of the reform with respect to the whole population.
Based on the household net incomes we estimate the distributional and the labour supply
e¤ects of the analysed tax reforms. For the econometric estimation of labour supply elasticities,
we apply a structural discrete choice household labour supply model following Van Soest (1995).
In the standard continuous model (see Hausman (1985)), labour supply responds along the intensive margin: an in…nitesimal change of the marginal tax rate changes the working hours only
a little, whereas participation responses cannot be satisfactorily analysed within this framework
(Blundell and MaCurdy (1999)). Discrete choice labour supply models allow to analyse both
the extensive (participation) and the intensive (hours worked) labour supply decision within
the same modelling framework (Blundell and MaCurdy (1999), Van Soest and Das (2001), and
Van Soest et al. (2002)). The intensive decision depends on the e¤ective marginal tax rate,
whereas the extensive participation decision depends on the tax wedge between gross (pre-tax)
labour costs and the after-tax net income of workers (see Kleven and Kreiner (2006)). The
continuous model “appears not to capture the data, in the sense that the number of part-time
jobs is strongly overpredicted”(Van Soest (1995)). There seems to be a lack of part-time jobs
because of …xed costs of hiring workers or increasing returns to scale of the worker’s production. Furthermore, because of …xed costs of working (Cogan (1981)), individuals are not willing
to work below a minimum number of hours. In addition, there are working time regulations
18
that limit the number of possible working hours to a discrete set. Therefore, a discrete choice
between distinct categories of working time seems to be more realistic than a continuum of
in…nitesimal choices. Using a discrete choice labour supply model has also the advantage to
model nonlinear budget constraints as a result of, for example, nonlinear taxes, joint …ling and
unemployment bene…ts (see MaCurdy et al. (1990), Van Soest (1995) or Blundell and MaCurdy
(1999)) and to incorporate a richer stochastic speci…cation in terms of unobserved wage rates
of nonworkers and random preferences.
Finally, FiFoSiM contains a CGE module for the estimation of growth and employment
e¤ects, which is linked to the tax bene…t module. This interaction allows for a better calibration of the model parameters and a more accurate estimation of the various e¤ects of reform
proposals. FiFoSiM has so far used either the top-down or the bottom-up approach to combine the microsimulation and the CGE module. In the bottom-up linkage the representative
household (income, labour supply, tax payments) in the CGE module is calibrated based on
the simulation results of the microsimulation modules. For the top-down linkage changes of the
wage or price level are computed in the CGE model and used in the microsimulation modules
for the calculation of net incomes and the labour supply estimation. The top-down bottom-up
approach used for this analysis is so far only executed manually and not automatically.27
5
Cash ‡ow ‡at tax proposal
The proposal of Mitschke (2004) in its original version combines an almost ‡at rate tax (two
brackets with di¤erent marginal rates) on earned income with a S-base cash ‡ow tax, i.e. income
which is invested in …rms is tax exempt.28 Therefore, the neutrality of the savings and investment
decision is achieved through this S-base cash ‡ow tax. In e¤ect, this reform proposal is a switch
from an income based tax system to consumption taxation (concept of deferred taxation).29
In principle, this proposal is close to the ”Flat Tax“ idea of Hall and Rabushka (1985) which
combines a R-base cash ‡ow taxation on corporate income with the same single marginal tax
rate on labour income.30 Essentially, the HR ‡at tax is a consumption-type, origin-based VAT
27
So far, the MSM module is written in Stata and the CGE module in GAMS. The interfacing and the
information exchange has to be executed manually. We are currently implementing a routine to automatically
execute the GAMS program from Stata, read in the modi…ed CGE parameters, resolve the MSM model, write
the modi…ed MSM in GAMS format, execute the CGE model and so on until convergence is achieved. However,
it appears to be useful to manually check the results to be able to quickly identify possible problems.
28
See Fuest et al. (2005a) and Fuest et al. (2007a) for a detailed description and analysis of the Mitschke
proposal.
29
See also Auerbach (2006).
30
See King (1987) and OECD (2007) for a review of the di¤erent concepts of corporate cash ‡ow taxes.
19
with a tax credit for labour income.31
In contrast to Mitschke (2004), who chooses a progressive tax schedule with two brackets, we
model a single marginal tax rate of 25%32 for all types of income with a basic allowance of 7,500
Euros in this paper. The Mitschke proposal further distinguishes between an introductory phase
(personal income tax reform) and a …nal phase (personal income tax and cash ‡ow corporate
tax). In the …rst phase, only the personal income tax system is changed to a system with a
single marginal rate on all sources of income (including capital and business income). In the
…nal phase, the modi…ed personal income tax is combined with a cash ‡ow corporate income
tax with the same marginal rate.33 Furthermore, an imputed rent on owner occupied housing
is also part of the tax base in this phase. For both phases the long-term revenue, employment
and growth e¤ects are simulated as well as the distributional e¤ects.
6
Analysis
In the …rst step, the …scal e¤ects are analysed in the tax bene…t module without taking into
account the behavioural reactions of the economic agents (…rst round e¤ects). In the second
step, we allow for behavioural reactions by estimating the labour supply responses (second
round e¤ects). We …nd considerable di¤erences in the labour supply reactions between couples
and singles as well as between men and women. While married men increase their labour supply
the strongest, single women even slightly decrease their labour supply. In the third step, the
labour demand and wage changes (third round e¤ects) are computed in the CGE module. In
the fourth step, the micro data information is used to calibrate the representative household
in the CGE module for the computation of the overall employment and growth e¤ects (general
equilibrium). We link the tax bene…t module to the CGE model by using the microsimulation
results to calibrate the representative household in terms of income, labour supply and tax
payments. As we use static MSM and CGE models, the behavioural adjustments are computed
in the long-run, whereas the …rst round e¤ects represent the immediate short-run e¤ects the
“day after”the reform. The main results are summarised in table 2.
The Mitschke proposal includes measures to broaden the tax base, therefore it is not clear
ex ante if the tax revenue will be higher or lower than in the status quo.34 The shift from the
31
See Keen et al. (2007). The tax base is sales minus purchases with capital goods being excluded (R-base).
Further on, this origin-based VAT is a tax on domestic production that taxes exports but not imports (in
contrast to the destination-based form of VAT).
32
The marginal rate of 25% is computed from micro data as an average tax rate of taxpayers under the
Mitschke proposal. The basic allowance is chosen such that the Mitschke ‡at tax yields the same revenue as
the Mitschke two bracket schedule.
33
Note that in contrast to HR, the VAT is not changed in the Mitschke proposal.
34
It would have been possible to construct the scenarios revenue neutral like in the previous sections. However,
20
Model Round E¤ect
PIT
MSM 1
Tax revenue
-2 billion e
2
Labour supply
+103,000
CGE
3
Labour demand
+370,000
Link
4
Tax revenue after adj. +3 billion e
4
Employment
+337,000
4
GDP
+1.1%
PIT + CIT
-13 billion e
+251,000
+540,000
-6 billion e
+471,000
+1.7%
Table 2: Summary of results for the HR type ‡at tax reform
current German tax regime to the Mitschke proposal would result in revenue losses amounting
to e 2 billion in the introductory phase (i.e. ‡at personal income tax, PIT) respectively e
13 billion in the …nal phase (i.e. ‡at personal income tax and cash ‡ow ‡at corporate income
tax, CIT) without taking the behavioural responses into account (…rst round e¤ects). These
short-run e¤ects indicate that the (not revenue neutral) ‡at tax reform scenario reduces the
average tax burden on labour. As a consequence of this lower tax wedge, the net wage is likely
to increase and the gross wage is likely to decrease. These e¤ects imply increasing labour supply
as well as increasing labour demand due to reduced user costs of labour. These e¤ects will be
simulated in the next step. As mentioned before, the behavioural responses are expectations
for the long-run (partial or general) equilibrium as both models are static, i.e. not taking the
transition path into account.
Labour supply would increase by 103,000 [251,000] fulltime equivalents. By including those
second round e¤ects, revenue increases and revenue losses are lowered. So far, these results are
only based on the MSM. This information is now used to calibrate the representative household
of the CGE model to derive the third round e¤ects: Labour demand would increase by 370,000
[540,000] due to reduced costs of capital and labour.35 Taking these e¤ects on wages and
prices into account allows us to resolve both models until they converge. This leads to the
following results: employment would grow by 337,000 full-time jobs, and GDP would increase
by 1.1% in the introductory phase. The overall employment e¤ects are larger than the labour
supply reactions because of reduced costs of labour and capital resulting in increasing labour
and investment demand.36 This result indicates the importance of taking general equilibrium
e¤ects into account. For the …nal phase, we calculate a total of 471,000 new full-time jobs and
a 1.7% increase in GDP. These results show that a cash ‡ow ‡at tax leads to further e¢ ciency
the HR idea as well as the Mitschke proposal are not designed to be revenue neutral. Furthermore, allowing for
a …rst round loss in revenue might trigger stronger e¢ ciency e¤ects than a revenue neutral scenario. Therefore,
the analysis in this section allows for a loss (or increase) in tax revenue.
35
This result implies that the incidence of the tax reform is split between employees and employers.
36
These results are in line with results from Aaberge et al. (2007) for Norway. They also derive for a
‡at personal income tax scenario that the general equilibrium e¤ects are larger than the pure labour supply
reactions.
21
gains due to more investment and labour demand as a consequence of reduced tax distortions
in the corporate sector.
Can such a reform overcome the fundamental equity e¢ ciency trade-o¤? The distributional
e¤ects before any (short run) and after the complete (long run) adjustment process are presented
in Table 3.
Decile
Short-run
PIT PIT+CIT
Long-run
PIT PIT+CIT
1
2
3
4
5
6
7
8
9
10
Gini
0,67
0,00
-0,29
-0,66
-1,27
-1,89
-2,31
-2,56
-2,27
2,11
1,16
77,93
11,68
6,33
3,30
0,73
-0,54
-2,03
-3,35
-3,24
1,69
-2,75
0,86
-0,28
-1,71
-2,54
-2,64
-2,53
-1,75
-0,31
0,73
5,92
4,48
78,18
11,41
4,70
1,47
-0,54
-1,00
-1,09
-0,68
0,95
6,47
1,28
Table 3: Changes in household disposable income (in percent)
Without taking any behavioural responses or adjustments into account (short-term …rst
round e¤ects) inequality increases in both phases. The main reason is the relief for the top of
the distribution. The small gains at the lower end cannot compensate the higher burden in the
middle income range and therefore inequality increases. These e¤ects change, however, after
the economy has fully equilibrated. The introduction of the personal income ‡at tax reduces
inequality because of the strong behavioural responses at the bottom of the distribution. When
combining the personal income ‡at tax with the corporate cash-‡ow tax, however, inequality
still increases but less than without behavioural adjustment. This is due to the fact that
especially the high income households have corporate or business income.
To sum up, taking the general equilibrium e¤ects into account has important implications
for the evaluation of tax reforms. The analysis shows that a personal income ‡at tax can
indeed overcome the fundamental equity e¢ ciency trade-o¤ in the long-run while simultaneously
increasing the tax revenue. However, this result does not hold for a cash ‡ow ‡at tax combining
a personal income ‡at tax with a corporate cash ‡ow ‡at tax, even when allowing for an ex-post
loss in revenue.
22
7
Conclusion
The success of the ‡at rate income tax in Eastern Europe suggests that this concept could
also be a model for countries of Western Europe. A linked MSM-CGE model provides a
powerful tool for the ex-ante evaluation of hypothetical tax bene…t reform proposals and is
the most appropriate method for a comprehensive analysis of the distributional, employment
and growth e¤ects of (‡at) tax reforms. When interpreting these results and especially the
e¢ ciency e¤ects, it has to be taken into account that we have limited our analysis to static
models. Therefore, the e¤ects from our analysis only account for the new long-run equilibrium
neglecting the transition path.37 However, regarding the political feasibility of a ‡at tax reform,
the short-term e¤ects are most likely to be decisive.38
In this paper, we analysed the introduction of a comprehensive cash ‡ow ‡at tax in the
tradition of Hall and Rabushka (1985). Using the linked MSM-CGE module, we were able
to account for the long-run general equilibrium e¤ects. The analysis showed that the overall
employment e¤ects are larger than the labour supply reactions because of reduced costs of
labour and capital resulting in increasing labour and investment demand. Therefore, it is
important to take these general equilibrium e¤ects into account. In doing so, the analysis
shows that a personal income ‡at tax can overcome the familiar equity e¢ ciency trade-o¤,
but only in the long-run. The adverse immediate distributional e¤ects still dominate in the
short-run. However, combining this ‡at tax with a cash ‡ow ‡at tax on business income
with the same marginal rate still increases inequality due to the large gains at the top of the
distribution at the expense of the middle class. This is important from a political economy
perspective. A strong and politically powerful middle class is a typical characteristic of most
Western European countries. This suggests that it will be hard for ‡at tax reforms to spill
over to these grown-up democracies. Since our analysis focuses on Germany, the question
arises whether the main …ndings are likely to apply to other countries as well. Therefore, more
(and especially comparative) country studies are required to complete the picture. However, a
multi-country linked MSM-CGE model has not been developed yet.
An aspect that is neglected in our analysis is the impact of tax reforms on training and
human capital accumulation. The results in Jacobs et al. (2007) suggest that ‡at tax reforms
may increase investment in skill formation and thus change the composition of the labour
37
Flat taxes are also supposed to have positive dynamic e¢ ciency and growth e¤ects (see e.g. Stokey and
Rebelo (1995) or Cassou and Lansing (2004)).
38
People tend to judge future gains and losses asymmetrically (see e.g. the “prospect theory“ by Kahneman
and Tversky (1979)). Starting from a reference point (status quo) and given the same variation in absolute
values, there is a bigger impact of losses than of gains (loss aversion). Furthermore, people prefer the status quo
over uncertain outcomes in the future (“status-quo-bias“, see Kahneman et al. (1991)). Therefore, short-term
losses in comparison to the status quo can have a much stronger impact than (possible) future gains. Hence,
the short term e¤ects presented here could be decisive.
23
force in the long term. But the question arises whether the income tax is the best instrument
to achieve this. Furthermore, our analysis abstracts from e¤ects of the ‡at tax reform on
compliance. Flat rate tax systems are widely expected to improve taxpayer compliance. The
2001 tax reform in Russia is widely thought to be an example for this e¤ect. Indeed, tax
compliance and revenue apparently improved by about one third after the 2001 tax reform
(Ivanova et al. (2005)). However, it is not clear whether this can be attributed solely to the ‡at
tax or to improved law enforcement and tax administration which was also part of the 2001
reform (see also Gaddy and Gale (2005) and Gorodnichenko et al. (2007)). Moreover, the case
of Russia di¤ers from Germany insofar as the latter has a long tradition of income taxation
in a market economy and a well established tax administration to ensure tax compliance. In
addition, since we do not change social insurance contributions, the marginal tax rate on labour
still remains high. This suggests that positive e¤ects of a ‡at tax reform on compliance are
probably less important in Germany than in the transition countries of Eastern Europe.
Furthermore, the question arises whether the scope of increasing growth and employment
through personal income tax reforms is su¢ ciently large. The user costs of labour and capital, which play an important role in determining the demand for labour and investment, are
rather determined by social security contributions and corporate taxes than by the personal
income tax. Including the CIT in the analysis does indeed lead to larger e¢ ciency e¤ects, but
at the expense of increasing inequality. Therefore, the main problem of implementing a ‡at
tax would be to convince a majority of the population that an immediate redistribution in
favour of the highest income deciles is acceptable to achieve (uncertain) future e¢ ciency gains.
Furthermore, it is uncertain whether a tax system that abolishes a large number of exemptions
and tax reliefs is politically sustainable. The temptation for politicians to serve special interest
groups with special deductions will not easily disappear. Moreover, from a political economy
perspective, a broad tax base allows the government to increase revenue with small increases
in tax rates. Therefore, narrow tax bases might protect the taxpayers from excess taxation by
the government.39
References
Aaberge, R., Colombino, U., Holmøy, E., Strøm, B. and Wennemo, T. (2007). Population
Ageing and Fiscal Sustainability: Integrating Detailed Labour Supply Models with CGE
Models, in A. Harding and A. Gupta (eds), Modelling Our Future: Social Security and
Taxation, Vol. I, Elsevier, Amsterdam, pp. 259–290.
39
Cf. Brennan and Buchanan (1980).
24
Aaberge, R., Colombino, U. and Strøm, S. (2000). Labor Supply Responses and Welfare E¤ects
from Replacing Current Tax Rules by a Flat Tax: Empirical Evidence from Italy, Norway
and Sweden, Journal of Population Economics 13: 595–621.
Altig, D., Auerbach, A., Kotliko¤, L., Smetters, K. and Walliser, J. (2001). Simulating fundamental tax reform in the U. S., American Economic Review 91(3): 574–595.
Anderson, J. M. (1990). Micro-Macro Linkages in Economic Models, in G. H. Lewis and R. C.
Michel (eds), Microsimulation Techniques for Tax and Transfer Analysis, Urban Institute
Press.
Arntz, M., Boeters, S. and Gürtzgen, N. (2006). Alternative Approaches to Discrete Working
Time Choice in an AGE Framework, Economic Modelling 23: 1008–1032.
Arntz, M., Boeters, S., Gürtzgen, N. and Schubert, S. (2006). Analysing Welfare Reform in a
Microsimulation-AGE Model: The Value of Disaggregation, ZEW Discussion Paper No.
06-076.
Auerbach, A. J. (2006). The Choice Between Income and Consumption Taxes: A Primer,
NBER Working Paper No. 12307.
Auerbach, A. and Kotliko¤, L. (1987). Dynamic Fiscal Policy, Cambridge University Press,
Cambridge.
Auerbach, A., Kotliko¤, L. and Skinner, J. (1983). The E¢ ciency Gains from Dynamic Tax
Reform, International Economic Review 24: 97 –130.
Ballard, C., Fullerton, D., Shoven, J. and Whalley, J. (1985). A General Equilibrium Model for
Tax Policy Evaluation, University of Chicago Press, Chicago.
Bergman, L. (2005). CGE Modeling of Environmental Policy and Resource Management, in
K. G. Mäler and J. R. Vincent (eds), Handbook of Environmental Economics, NorthHolland, Amsterdam, chapter 24, pp. 1273–1306.
Bergs, C. and Peichl, A. (2008). Numerische Gleichgewichtsmodelle zur Analyse von Politikreformen, Zeitschrift für Wirtschaftspolitik 57 (1): 1–26.
Betson, D. (1990). How Reliable are Conclusions Derived from Microsimulation Models?, in
J. Brunner and H.-G. Petersen (eds), Simulation Models in Tax and Transfer Policy:
Proceedings of an International Symposium, Campus Verlag, Frankfurt am Main, pp. 423–
446.
25
Blanchard, O. (1985). Debt, De…cits, and Finite Horizons, Journal of Political Economy
93(2): 223–247.
Blundell, R. and MaCurdy, T. (1999). Labor Supply: A Review of Alternative Approaches,
in O. Ashenfelter and D. Card (eds), Handbook of Labor Economics, Vol. 3A, Elsevier,
Amsterdam, pp. 1559–1695.
Boeters, S., Böhringer, C., Büttner, T. and Kraus, M. (2006). Economic E¤ects of VAT Reform
in Germany, ZEW Discussion Paper No. 06-030.
Boeters, S., Feil, M. and Gürtzgen, N. (2005). Discrete Working Time Choice in an Applied
General Equilibrium Model, Computational Economics 26: 183–211.
Boeters, S., Gürtzgen, N. and Schnabel, R. (2006). Reforming Social Welfare in Germany - An
Applied General Equilibrium Analysis, German Economic Review 7: 363–388.
Böhringer, C., Boeters, S. and Feil, M. (2005). Taxation and Unemployment: An Applied
General Equilibrium Approach, Economic Modelling 22: 81–108.
Böhringer, C. and Rutherford, T. (2006). Combining Top-Down and Bottom-up in Energy
Policy Analysis: A Decomposition Approach, ZEW Discussion Paper No. 06-007.
Bourguignon, F., Robilliard, A.-S. and Robinson, S. (2003). Representative versus Real
Households in the Macro-Economic Modelling of Inequality, DIAL Document de Travail DT/2003-10.
Bourguignon, F. and Spadaro, A. (2006). Microsimulation as a tool for evaluating redistribution
policies, Journal of Economic Inequality 4(1): 77–106.
Bovenberg, L. (2003). Tax Policy and Labor Market Performance, CESifo Working Paper No.
1035.
Brennan, G. and Buchanan, J. (1980). The Power to Tax: Analytical Foundations of a Fiscal
Constitution, Cambridge University Press, New York.
Browning, E. and Browning, J. (1985). Why Not a True Flat Rate Tax, Cato Journal 5: 127–
143.
Cajner, T., Grobovsek, J. and Kozamernik, D. (2006). Welfare and E¢ ciency E¤ects of Alternative Tax Reforms in Slovenia, IB Revija (IB Review) XXXX (1-2): 125–136.
26
Caminada, K. and Goudswaard, K. (2001). Does a Flat Rate Individual Income Tax Reduce
Tax Progressivity? A Simulation for the Netherlands, Public Finance and Management 1
(4): 471–499.
Cassou, S. P. and Lansing, K. J. (2004). Growth E¤ects of Shifting from a Graduated-rate Tax
System to a Flat Tax, Economic Inquiry 42(2): 194–213.
Cogan, J. (1981). Fixed costs and labor supply, Econometrica 49: 945–963.
Cogneau, D. and Robilliard, A.-S. (2000). Growth, Distribution and Poverty in Madagascar:
Learning from a Microsimulation Model in a General Equilibrium Framework, International Food Policy Research Institute TMD Discussion Paper 61.
Conrad, K. (1991). Intertemporal General Equilibrium Modelling of German Tax Reform
Proposals, Finanzarchiv 48 (2): 194–221.
Cororaton, C. B., Cockburn, J. and Corong, E. (2005). Doha Scenarios, Trade Reforms, and
Poverty in the Philippines: A Computable General Equilibrium Analysis, World Bank
Policy Research Working Paper 3738.
Creedy, J., Duncan, A., Harris, M. and Scutella, R. (2002). Microsimulation Modelling of
Taxation and the Labour Market: the Melbourne Institute Tax and Transfer Simulator,
Edward Elgar Publishing, Cheltenham.
Davies, J. (2004). Microsimulation, CGE and Macro Modelling for Transition and Developing
Economies, Mimeo, University of Western Ontario.
Díaz Giménez, J. and Pijoan-Mas, J. (2006). Flat Tax Reforms in the US: A Boon for the
Income Poor, CEPR Discussion Paper No. 5812.
Decoster, A. and Orsini, K. (2007). Verdient een vlaktaks zichzelf terug?, Leuvense Economische
Standpunten, K.U. Leuven, Centrum voor Economische Studiën.
Dunbar, A. and Pogue, T. (1998). Estimating Flat Tax Incidence and Yield: A Sensitivity
Analysis, National Tax Journal 51: 303–324.
Fehr, H. and Wiegard, W. (1996). Numerische Gleichgewichtsmodelle: Grundstruktur, Anwendungen und Erkenntnisgehalt, Jahrbuch 13: Experimente in der Ökonomie, Ökonomie
und Gesellschaft, Frankfurt, pp. 296–339.
Fehr, H. and Wiegard, W. (1998). German Income Tax Reforms: Separating E¢ ciency from
Redistribution, in A. Fossati and J. Hutton (eds), Policy Simulations in the European
Union, Routledge, London, pp. 235–263.
27
Fuest, C., Peichl, A. and Schaefer, T. (2005a). Aufkommens-, Beschäftigungs- und Wachstumswirkungen einer Steuerreform nach dem Vorschlag von Mitschke, FiFo-Bericht 05-2005.
Fuest, C., Peichl, A. and Schaefer, T. (2005b). Dokumentation FiFoSiM: Integriertes
Steuer-Transfer-Mikrosimulations- und CGE-Modell, Finanzwissenschaftliche Diskussionsbeiträge Nr. 05 - 03.
Fuest, C., Peichl, A. and Schaefer, T. (2007a). Aufkommens-, Beschäftigungs- und Wachstumswirkungen einer Steuerreform nach dem Vorschlag von Mitschke, Nomos, Baden-Baden.
Fuest, C., Peichl, A. and Schaefer, T. (2007b). Is a Flat Tax politically feasible in a grown-up
Welfare State?, CPE discussion paper 07-6, University of Cologne.
Gaddy, C. G. and Gale, W. G. (2005). Demythologizing the Russian Flat Tax, Tax Notes
International 43: 983–988.
Gale, W. G., Houser, S. and Scholz, J. K. (1996). Distributional E¤ects of Fundamental Tax
Reform, in H. J. Aaron and W. G. Gale (eds), Economic E¤ects of Fundamental Tax
Reform, The Brookings Institution, Washington, D. C., pp. 281–320.
González-Torrabadella, M. and Pijoan-Mas, J. (2006). Flat tax reforms: a general equilibrium
evaluation for Spain, Investigaciones Económicas XXX (2): 317–351.
Gorodnichenko, Y., Martinez-Vazquez, J. and Peter, K. S. (2007). Myth and Reality of Flat
Tax Reform: Micro Estimates of Tax Evasion Response and Welfare E¤ects in Russia,
IZA Discussion Paper No. 3267.
Graa‡and, J., de Mooij, R. and Nibbelink, A. (2001). MIMICing Tax Policies and the Labour
Market, Elsevier, Amsterdam.
Gupta, A. and Kapur, V. (2000). Microsimulation in Government Policy and Forecasting,
North-Holland, Amsterdam.
Haan, P. and Steiner, V. (2005). Distributional E¤ects of the German Tax Reform 2000 - A
Behavioral Microsimulation Analysis, Schmollers Jahrbuch 125 (1): 39–49.
Hall, R. E. and Rabushka, A. (1985). The Flat Tax, Hoover Institution Press, Stanford.
Harding, A. (1996). Microsimulation and public policy, North-Holland, Elsevier, Amsterdam.
Harris, R. (1984). Applied General Equilibrium Analysis of Small Open Economies with Scale
Economies and Imperfect Competition, American Economic Review 74: 1017–1032.
28
Hausman, J. (1985). Taxes and Labor Supply, in A. Auerbach and M. Feldstein (eds), Handbook
of Public Economics, North-Holland, Amsterdam, pp. 213–263.
Heckman, J. (1993). What has been learned about labor supply in the past twenty years?,
American Economic Review Papers and Proceedings 85: 116–121.
Heer, B. and Trede, M. (2003). E¢ ciency and distribution e¤ects of a revenue-neutral income
tax reform, Journal of Macroeconomics 25: 87–107.
Ho, M. and Stiroh, K. (1998). Revenue, Progressivity and the Flat Tax, Contemporary Economic Policy XVI: 85–97.
Hérault, N. (2007). Trade Liberalisation, Poverty and Inequality in South Africa: A CGEMicrosimulation analysis, Economic Record 83 (262): 317–328.
Hutton, J. and Ruocco, A. (1999). Tax Reform and Employment in Europe, International Tax
and Public Finance 6: 263–287.
Immervoll, H., Kleven, H., Kreiner, C. and Saez, E. (2007). Welfare Reform in European
Countries: A Micro-Simulation Analysis, The Economic Journal 117 (516): 1–44.
Ivanova, A., Keen, M. and Klemm, A. (2005). Russia’s ’‡at tax’, Economic Policy July: 397–
444.
Jacobs, B., de Mooij, R. A. and Folmer, K. (2007). Analyzing a ‡at income tax in the Netherlands, Tinbergen Institute Discussion Paper 2007-029/3.
Johansen, L. (1960). A Multi-sectoral Study of Economic Growth, North-Holland, Amsterdam.
Kahneman, D., Knetsch, J. L. and Thaler, R. H. (1991). Anomalies: The endowment e¤ect,
loss aversion, and status quo bias, Journal of Economic Perspectives 5: 193–206.
Kahneman, D. and Tversky, A. (1979). Prospect theory: An analysis of decision under risk,
Econometrica 47: 263–291.
Keen, M., Kim, Y. and Varsano, R. (2007). The ’‡at tax(es)’: Principles and experience,
International Tax and Public Finance forthcoming.
Kehoe, P. and Kehoe, T. (1994). A Primer on Static Applied General Equilibrium Models,
Federal Reserve Bank of Minneapolis Quaterly Review 18(1).
Kehoe, T. (1998). Social Accounting Matrices and Applied General Equilibrium Models, Applied
Economics and Public Policy pp. 59–87.
29
Kehoe, T. and Prescott, E. (1995). Introduction to the Symposium: The Discipline of Applied
General Equilibrium, Economic Theory 6: 1–11.
Keuschnigg, C. and Dietz, M. (2003). Unternehmenssteuerreform II - quantitative Auswirkungen auf Wachstum und Verteilung, Paul Haupt, Bern.
King, M. (1987). The Cash Flow Corporate Income Tax, in M. Feldstein (ed.), The E¤ects of
Taxation on Capital Accumulation, University of Chicago Press, Chicago.
Kleven, H. J. and Kreiner, C. (2006). The Marginal Cost of Public Funds in OECD Countries:
Hours of Work versus Labor Force Participation, Journal of Public Economics 90: 1955–
1973.
Lau, L. (1984). Comments on Mansur and Whalley (1984), in H. Scarf and J. B. Shoven (eds),
Applied General Equilibrium Analysis, Cambridge University Press, New York, pp. 127–
137.
MaCurdy, T., Green, D. and Paarsch, H. (1990). Assessing Empirical Approaches for Analyzing
Taxes and Labor Supply, Journal of Human Resources 25(3): 415–490.
Mansur, A. and Whalley, J. (1984). Numerical Speci…cation of Applied General Equilibrium
Models: Estimation, Calibration and Data, in H. Scarf and J. B. Shoven (eds), Applied
General Equilibrium Analysis, Cambridge University Press, New York, pp. 69–127.
Merz, J., Stolze, H. and Zwick, M. (2002). Professions, entrepreneurs, employees and the
new German tax (cut) reform 2000. A MICSIM microsimulation analysis of distributional
impacts, FFB-Diskussionspapier Nr. 35, Universität Lüneburg.
Merz, J. and Zwick, M. (2002). Verteilungswirkungen der Steuerreform 2000/2005 im Vergleich
zum Karlsruher Entwurf. Auswirkungen auf die Einkommensverteilung bei Selbständigen,
Wirtschaft und Statistik 8: 729–740.
Mitschke, J. (2004). Erneuerung des deutschen Einkommensteuerrechts: Gesetzestextentwurf
und Begründung, Verlag Otto Schmidt, Köln.
Nielsen, S. B., Frederiksen, N. K. and Lassen, D. D. (1999). Would the ‡at tax fall ‡at in
Denmark?, in T. M. Andersen, S. E. H. Jensen and O. Risager (eds), Macroeconomic
Perspectives on the Danish Economy, MacMillan, pp. 106–133.
OECD (2007). Fundamental Reform of Corporate Income Tax, OECD Tax Policy Studies 16.
30
O’Hare, J. and Gupta, A. (2000). Practical Aspects of Microsimulation Modelling, in A. Gupta
and V. Kapur (eds), Microsimulation in Government Policy and Forecasting, NorthHolland, Amsterdam, pp. 563–640.
Orcutt, G. (1957). A new type of socio-economic system, Review of Economics and Statistics
58: 773–797.
Orcutt, G. (1967). Microeconomic Analysis for Prediction of National Accounts, in H. Wold,
G. H. Orcutt, E. A. Robinson, D. Suits and P. de Wol¤ (eds), Forecasting on a Scienti…c
Basis - Proceedings of an International Summer Institute, Gulbenkian Foundation, Lisbon.
Orcutt, G., Greenberg, M., Korbel, J. and Rivlin, A. (1961). Microanalyis of Socioeconomic
Systems: A Simulation Study, Harper and Row, New York.
Paulus, A. and Peichl, A. (2008). E¤ects of ‡at tax reforms in Western Europe on equity and
e¢ ciency, EUROMOD Working Paper No. EM2/08, University of Essex.
Peichl, A. (2005). Die Evaluation von Steuerreformen durch Simulationsmodelle, Finanzwissenschaftliche Diskussionsbeiträge Nr. 05-01.
Peichl, A. and Schaefer, T. (2006). Documentation FiFoSiM: Integrated Tax Bene…t Microsimulation and CGE Model, Finanzwissenschaftliche Diskussionsbeiträge Nr. 06 - 10, Universität zu Köln.
Pyatt, G. and Round, J. (1985). Social Accounting Matrices. A Basis for Planning, The World
Bank, Washington D.C.
Radulescu, D. and Stimmelmayr, M. (2004). Implementing a Dual Income Tax in Germany:
E¤ects on Investment and Welfare, Mimeo, Universität München.
Ramsey, F. (1928). A Mathematical Theory of Saving, Economic Journal 543-559: 28.
Rausch, S. and Rutherford, T. F. (2007). Computation of Equilibria in OLG Models With
Many Heterogeneous Households, Ruhr Economic Paper No. 15.
Rutherford, T., Tarr, D. and Shepotylo, O. (2005). Poverty E¤ects of Russia’s WTO Accession:
Modelling ‘Real’ Households and Endogenous Productivity E¤ects, World Bank Policy
Research Working Paper 3473.
Savard, L. (2003). Poverty and Income Distribution in a CGE-Household Micro- Simulation
Model: Top-Down/Bottom Up Approach, CIRPÉE Centre interuniversitaire sur le risque,
les politiques économiques et l’emploi. Working Paper 03-43.
31
Shoven, J. and Whalley, J. (1984). Applied General Equilibrium Models of Taxation and International Trade: An Introduction and Survey, Journal of Economic Literature 22: 1007–
1051.
Shoven, J. and Whalley, J. (1992). Applying General Equilibrium, Cambridge University Press,
Cambridge.
Sørensen, P. (2002). The German Business Tax Reform of 2000 - A General Equilibrium
Analysis, German Economic Review 3: 347–378.
Steiner, V. and Wrohlich, K. (2004). Household Taxation, Income Splitting and Labor Supply
Incentives - A Microsimulation Study for Germany, CESifo Economic Studies 50(3): 541–
568.
Stokey, N. L. and Rebelo, S. (1995). Growth E¤ects of Flat-Rate Taxes, Journal of Political
Economy 103(3): 519–550.
Van Soest, A. (1995). Structural Models of Family Labor Supply: A Discrete Choice Approach,
Journal of Human Resources 30: 63–88.
Van Soest, A. and Das, M. (2001). Family Labor Supply and Proposed Tax Reforms in the
Netherlands, De Economist 149(2): 191–218.
Van Soest, A., Das, M. and Gong, X. (2002). A Structural Labour Supply Model with ‡exible
Preferences, Journal of Econometrics 107: 345 –374.
Ventura, G. (1999). Flat Tax Reform: A Quantitative Exploration, Journal of Economic
Dynamics and Control 23: 1425–1458.
Verboven, F. (1996). The nested logit model and representative consumer theory, Economic
Letters 50 (1): 57–63.
Wagenhals, G. (2001). Incentive and Redistribution E¤ects of the German Tax Reform 2000,
Finanzarchiv 57: 316–332.
Wagenhals, G. (2004). Tax-bene…t microsimulation models for Germany: A Survey, IAWReport / Institut fuer Angewandte Wirtschaftsforschung (Tübingen) 32(1): 55–74.
32
Download